Enforce compression extensions for CSV Files

src/planner/binder/statement/bind_copy.cpp

@@ -68,7 +68,6 @@ BoundStatement Binder::BindCopyTo(CopyStatement &stmt) {
 
 	auto original_options = stmt.info->options;
 	stmt.info->options.clear();
-
 	for (auto &option : original_options) {
 		auto loption = StringUtil::Lower(option.first);
 		if (loption == "use_tmp_file") {
@@ -105,6 +104,21 @@ BoundStatement Binder::BindCopyTo(CopyStatement &stmt) {
 			auto converted = ConvertVectorToValue(std::move(option.second));
 			partition_cols = ParseColumnsOrdered(converted, select_node.names, loption);
 		} else {
+			if (loption == "compression") {
+				if (option.second.empty()) {
+					// This can't be empty
+					throw BinderException("COMPRESSION option, in the file scanner, can't be empty. It should be set "
+					                      "to AUTO, UNCOMPRESSED, GZIP, SNAPPY or ZSTD. Depending on the file format.");
+				}
+				auto parameter = StringUtil::Lower(option.second[0].ToString());
+				if (parameter == "gzip" && !StringUtil::EndsWith(bind_input.file_extension, ".gz")) {
+					// We just add .gz
+					bind_input.file_extension += ".gz";
+				} else if (parameter == "zstd" && !StringUtil::EndsWith(bind_input.file_extension, ".zst")) {
+					// We just add .zst
+					bind_input.file_extension += ".zst";
+				}
+			}
 			stmt.info->options[option.first] = option.second;
 		}
 	}

test/sql/copy/csv/test_partition_compression.test

@@ -0,0 +1,63 @@
+# name: test/sql/copy/csv/test_partition_compression.test
+# description: Test we can round-trip partitioned compressed CSV files
+# group: [csv]
+
+statement ok
+PRAGMA enable_verification
+
+require no_extension_autoloading
+
+statement ok
+CREATE TABLE test AS VALUES ('a', 'foo', 1), ('a', 'foo', 2), ('a', 'bar', 1), ('b', 'bar', 1);
+
+statement ok
+COPY (FROM test) TO '__TEST_DIR__/data.csv.d' (FORMAT 'csv', COMPRESSION 'gzip', PARTITION_BY ('col0', 'col1'));
+
+# Specify Compression
+statement error
+FROM read_csv_auto('__TEST_DIR__/data.csv.d/*/*/*.csv')
+----
+No files found that match the pattern
+
+query III
+FROM read_csv_auto('__TEST_DIR__/data.csv.d/*/*/*.csv.gz');
+----
+a	bar	1
+a	foo	1
+a	foo	2
+b	bar	1
+
+query III
+FROM read_csv_auto('__TEST_DIR__/data.csv.d/*/*/*.csv.*');
+----
+a	bar	1
+a	foo	1
+a	foo	2
+b	bar	1
+
+require parquet
+
+statement ok
+COPY (FROM test) TO '__TEST_DIR__/data.csv.d2' (FORMAT 'csv', COMPRESSION 'zstd', PARTITION_BY ('col0', 'col1'));
+
+# Specify Compression
+statement error
+FROM read_csv_auto('__TEST_DIR__/data.csv.d2/*/*/*.csv')
+----
+No files found that match the pattern
+
+query III
+FROM read_csv_auto('__TEST_DIR__/data.csv.d2/*/*/*.csv.zst');
+----
+a	bar	1
+a	foo	1
+a	foo	2
+b	bar	1
+
+query III
+FROM read_csv_auto('__TEST_DIR__/data.csv.d2/*/*/*.csv.*');
+----
+a	bar	1
+a	foo	1
+a	foo	2
+b	bar	1